/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Robert Collins
  • Date: 2009-08-24 21:05:09 UTC
  • mto: This revision was merged to the branch mainline in revision 4645.
  • Revision ID: robertc@robertcollins.net-20090824210509-pproia2q9evq1nsl
lsprof support.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
from __future__ import absolute_import
18
 
 
19
17
import difflib
20
18
import os
21
19
import re
22
 
import string
 
20
import shutil
23
21
import sys
24
22
 
25
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
26
24
lazy_import(globals(), """
27
25
import errno
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
32
 
    cmdline,
33
 
    controldir,
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
34
34
    errors,
35
35
    osutils,
36
36
    patiencediff,
38
38
    timestamp,
39
39
    views,
40
40
    )
41
 
 
42
 
from breezy.workingtree import WorkingTree
43
 
from breezy.i18n import gettext
44
41
""")
45
42
 
46
 
from .registry import (
47
 
    Registry,
 
43
from bzrlib.symbol_versioning import (
 
44
    deprecated_function,
48
45
    )
49
 
from .sixish import text_type
50
 
from .trace import mutter, note, warning
51
 
from .tree import FileTimestampUnavailable
52
 
 
53
 
 
54
 
DEFAULT_CONTEXT_AMOUNT = 3
55
 
 
56
 
 
57
 
class AtTemplate(string.Template):
58
 
    """Templating class that uses @ instead of $."""
59
 
 
60
 
    delimiter = '@'
 
46
from bzrlib.trace import mutter, note, warning
61
47
 
62
48
 
63
49
# TODO: Rather than building a changeset object, we should probably
74
60
        self.opcodes = None
75
61
 
76
62
 
77
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
63
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
78
64
                  allow_binary=False, sequence_matcher=None,
79
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
65
                  path_encoding='utf8'):
80
66
    # FIXME: difflib is wrong if there is no trailing newline.
81
67
    # The syntax used by patch seems to be "\ No newline at
82
68
    # end of file" following the last diff line from that
87
73
    # In the meantime we at least make sure the patch isn't
88
74
    # mangled.
89
75
 
 
76
 
 
77
    # Special workaround for Python2.3, where difflib fails if
 
78
    # both sequences are empty.
 
79
    if not oldlines and not newlines:
 
80
        return
 
81
 
90
82
    if allow_binary is False:
91
83
        textfile.check_text_lines(oldlines)
92
84
        textfile.check_text_lines(newlines)
93
85
 
94
86
    if sequence_matcher is None:
95
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
96
 
    ud = patiencediff.unified_diff_bytes(oldlines, newlines,
97
 
                                         fromfile=old_label.encode(
98
 
                                             path_encoding, 'replace'),
99
 
                                         tofile=new_label.encode(
100
 
                                             path_encoding, 'replace'),
101
 
                                         n=context_lines, sequencematcher=sequence_matcher)
 
88
    ud = patiencediff.unified_diff(oldlines, newlines,
 
89
                      fromfile=old_filename.encode(path_encoding),
 
90
                      tofile=new_filename.encode(path_encoding),
 
91
                      sequencematcher=sequence_matcher)
102
92
 
103
93
    ud = list(ud)
104
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
94
    if len(ud) == 0: # Identical contents, nothing to do
105
95
        return
106
96
    # work-around for difflib being too smart for its own good
107
97
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
108
98
    if not oldlines:
109
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
99
        ud[2] = ud[2].replace('-1,0', '-0,0')
110
100
    elif not newlines:
111
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
101
        ud[2] = ud[2].replace('+1,0', '+0,0')
112
102
 
113
103
    for line in ud:
114
104
        to_file.write(line)
115
 
        if not line.endswith(b'\n'):
116
 
            to_file.write(b"\n\\ No newline at end of file\n")
117
 
    to_file.write(b'\n')
 
105
        if not line.endswith('\n'):
 
106
            to_file.write("\n\\ No newline at end of file\n")
 
107
    to_file.write('\n')
118
108
 
119
109
 
120
110
def _spawn_external_diff(diffcmd, capture_errors=True):
121
 
    """Spawn the external diff process, and return the child handle.
 
111
    """Spawn the externall diff process, and return the child handle.
122
112
 
123
113
    :param diffcmd: The command list to spawn
124
114
    :param capture_errors: Capture stderr as well as setting LANG=C
146
136
                                stdout=subprocess.PIPE,
147
137
                                stderr=stderr,
148
138
                                env=env)
149
 
    except OSError as e:
 
139
    except OSError, e:
150
140
        if e.errno == errno.ENOENT:
151
141
            raise errors.NoDiff(str(e))
152
142
        raise
154
144
    return pipe
155
145
 
156
146
 
157
 
# diff style options as of GNU diff v3.2
158
 
style_option_list = ['-c', '-C', '--context',
159
 
                     '-e', '--ed',
160
 
                     '-f', '--forward-ed',
161
 
                     '-q', '--brief',
162
 
                     '--normal',
163
 
                     '-n', '--rcs',
164
 
                     '-u', '-U', '--unified',
165
 
                     '-y', '--side-by-side',
166
 
                     '-D', '--ifdef']
167
 
 
168
 
 
169
 
def default_style_unified(diff_opts):
170
 
    """Default to unified diff style if alternative not specified in diff_opts.
171
 
 
172
 
        diff only allows one style to be specified; they don't override.
173
 
        Note that some of these take optargs, and the optargs can be
174
 
        directly appended to the options.
175
 
        This is only an approximate parser; it doesn't properly understand
176
 
        the grammar.
177
 
 
178
 
    :param diff_opts: List of options for external (GNU) diff.
179
 
    :return: List of options with default style=='unified'.
180
 
    """
181
 
    for s in style_option_list:
182
 
        for j in diff_opts:
183
 
            if j.startswith(s):
184
 
                break
185
 
        else:
186
 
            continue
187
 
        break
188
 
    else:
189
 
        diff_opts.append('-u')
190
 
    return diff_opts
191
 
 
192
 
 
193
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
147
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
194
148
                  diff_opts):
195
149
    """Display a diff by calling out to the external diff program."""
196
150
    # make sure our own output is properly ordered before the diff
197
151
    to_file.flush()
198
152
 
199
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
200
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
153
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
154
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
201
155
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
202
156
    newtmpf = os.fdopen(newtmp_fd, 'wb')
203
157
 
220
174
        if sys.platform == 'win32':
221
175
            # Popen doesn't do the proper encoding for external commands
222
176
            # Since we are dealing with an ANSI api, use mbcs encoding
223
 
            old_label = old_label.encode('mbcs')
224
 
            new_label = new_label.encode('mbcs')
 
177
            old_filename = old_filename.encode('mbcs')
 
178
            new_filename = new_filename.encode('mbcs')
225
179
        diffcmd = ['diff',
226
 
                   '--label', old_label,
 
180
                   '--label', old_filename,
227
181
                   old_abspath,
228
 
                   '--label', new_label,
 
182
                   '--label', new_filename,
229
183
                   new_abspath,
230
184
                   '--binary',
231
 
                   ]
 
185
                  ]
232
186
 
233
 
        diff_opts = default_style_unified(diff_opts)
 
187
        # diff only allows one style to be specified; they don't override.
 
188
        # note that some of these take optargs, and the optargs can be
 
189
        # directly appended to the options.
 
190
        # this is only an approximate parser; it doesn't properly understand
 
191
        # the grammar.
 
192
        for s in ['-c', '-u', '-C', '-U',
 
193
                  '-e', '--ed',
 
194
                  '-q', '--brief',
 
195
                  '--normal',
 
196
                  '-n', '--rcs',
 
197
                  '-y', '--side-by-side',
 
198
                  '-D', '--ifdef']:
 
199
            for j in diff_opts:
 
200
                if j.startswith(s):
 
201
                    break
 
202
            else:
 
203
                continue
 
204
            break
 
205
        else:
 
206
            diffcmd.append('-u')
234
207
 
235
208
        if diff_opts:
236
209
            diffcmd.extend(diff_opts)
237
210
 
238
211
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
239
 
        out, err = pipe.communicate()
 
212
        out,err = pipe.communicate()
240
213
        rc = pipe.returncode
241
214
 
242
215
        # internal_diff() adds a trailing newline, add one here for consistency
243
 
        out += b'\n'
 
216
        out += '\n'
244
217
        if rc == 2:
245
218
            # 'diff' gives retcode == 2 for all sorts of errors
246
219
            # one of those is 'Binary files differ'.
253
226
            out, err = pipe.communicate()
254
227
 
255
228
            # Write out the new i18n diff response
256
 
            to_file.write(out + b'\n')
 
229
            to_file.write(out+'\n')
257
230
            if pipe.returncode != 2:
258
231
                raise errors.BzrError(
259
 
                    'external diff failed with exit code 2'
260
 
                    ' when run with LANG=C and LC_ALL=C,'
261
 
                    ' but not when run natively: %r' % (diffcmd,))
 
232
                               'external diff failed with exit code 2'
 
233
                               ' when run with LANG=C and LC_ALL=C,'
 
234
                               ' but not when run natively: %r' % (diffcmd,))
262
235
 
263
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
236
            first_line = lang_c_out.split('\n', 1)[0]
264
237
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
265
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
238
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
266
239
            if m is None:
267
240
                raise errors.BzrError('external diff failed with exit code 2;'
268
241
                                      ' command: %r' % (diffcmd,))
281
254
                msg = 'exit code %d' % rc
282
255
 
283
256
            raise errors.BzrError('external diff failed with %s; command: %r'
284
 
                                  % (msg, diffcmd))
 
257
                                  % (rc, diffcmd))
 
258
 
285
259
 
286
260
    finally:
287
261
        oldtmpf.close()                 # and delete
288
262
        newtmpf.close()
289
 
 
290
 
        def cleanup(path):
291
 
            # Warn in case the file couldn't be deleted (in case windows still
292
 
            # holds the file open, but not if the files have already been
293
 
            # deleted)
294
 
            try:
295
 
                os.remove(path)
296
 
            except OSError as e:
297
 
                if e.errno not in (errno.ENOENT,):
298
 
                    warning('Failed to delete temporary file: %s %s', path, e)
299
 
 
300
 
        cleanup(old_abspath)
301
 
        cleanup(new_abspath)
302
 
 
303
 
 
304
 
def get_trees_and_branches_to_diff_locked(
305
 
        path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
263
        # Clean up. Warn in case the files couldn't be deleted
 
264
        # (in case windows still holds the file open, but not
 
265
        # if the files have already been deleted)
 
266
        try:
 
267
            os.remove(old_abspath)
 
268
        except OSError, e:
 
269
            if e.errno not in (errno.ENOENT,):
 
270
                warning('Failed to delete temporary file: %s %s',
 
271
                        old_abspath, e)
 
272
        try:
 
273
            os.remove(new_abspath)
 
274
        except OSError:
 
275
            if e.errno not in (errno.ENOENT,):
 
276
                warning('Failed to delete temporary file: %s %s',
 
277
                        new_abspath, e)
 
278
 
 
279
 
 
280
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
 
281
    apply_view=True):
306
282
    """Get the trees and specific files to diff given a list of paths.
307
283
 
308
284
    This method works out the trees to be diff'ed and the files of
319
295
    :param new_url:
320
296
        The url of the new branch or tree. If None, the tree to use is
321
297
        taken from the first path, if any, or the current working tree.
322
 
    :param add_cleanup:
323
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
324
 
        will register cleanups that must be run to unlock the trees, etc.
325
298
    :param apply_view:
326
299
        if True and a view is set, apply the view or check that the paths
327
300
        are within it
328
301
    :returns:
329
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
330
 
        specific_files, extra_trees) where extra_trees is a sequence of
331
 
        additional trees to search in for file-ids.  The trees and branches
332
 
        will be read-locked until the cleanups registered via the add_cleanup
333
 
        param are run.
 
302
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
303
        extra_trees is a sequence of additional trees to search in for
 
304
        file-ids.
334
305
    """
335
306
    # Get the old and new revision specs
336
307
    old_revision_spec = None
359
330
        default_location = path_list[0]
360
331
        other_paths = path_list[1:]
361
332
 
362
 
    def lock_tree_or_branch(wt, br):
363
 
        if wt is not None:
364
 
            wt.lock_read()
365
 
            add_cleanup(wt.unlock)
366
 
        elif br is not None:
367
 
            br.lock_read()
368
 
            add_cleanup(br.unlock)
369
 
 
370
333
    # Get the old location
371
334
    specific_files = []
372
335
    if old_url is None:
373
336
        old_url = default_location
374
337
    working_tree, branch, relpath = \
375
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
 
    lock_tree_or_branch(working_tree, branch)
 
338
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
377
339
    if consider_relpath and relpath != '':
378
340
        if working_tree is not None and apply_view:
379
341
            views.check_path_in_view(working_tree, relpath)
380
342
        specific_files.append(relpath)
381
343
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
382
 
    old_branch = branch
383
344
 
384
345
    # Get the new location
385
346
    if new_url is None:
386
347
        new_url = default_location
387
348
    if new_url != old_url:
388
349
        working_tree, branch, relpath = \
389
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
 
        lock_tree_or_branch(working_tree, branch)
 
350
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
391
351
        if consider_relpath and relpath != '':
392
352
            if working_tree is not None and apply_view:
393
353
                views.check_path_in_view(working_tree, relpath)
394
354
            specific_files.append(relpath)
395
355
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
396
 
                                 basis_is_default=working_tree is None)
397
 
    new_branch = branch
 
356
        basis_is_default=working_tree is None)
398
357
 
399
358
    # Get the specific files (all files is None, no files is [])
400
359
    if make_paths_wt_relative and working_tree is not None:
401
 
        other_paths = working_tree.safe_relpath_files(
402
 
            other_paths,
 
360
        try:
 
361
            from bzrlib.builtins import safe_relpath_files
 
362
            other_paths = safe_relpath_files(working_tree, other_paths,
403
363
            apply_view=apply_view)
 
364
        except errors.FileInWrongBranch:
 
365
            raise errors.BzrCommandError("Files are in different branches")
404
366
    specific_files.extend(other_paths)
405
367
    if len(specific_files) == 0:
406
368
        specific_files = None
407
 
        if (working_tree is not None and working_tree.supports_views() and
408
 
                apply_view):
 
369
        if (working_tree is not None and working_tree.supports_views()
 
370
            and apply_view):
409
371
            view_files = working_tree.views.lookup_view()
410
372
            if view_files:
411
373
                specific_files = view_files
412
374
                view_str = views.view_display_str(view_files)
413
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
375
                note("*** Ignoring files outside view. View is %s" % view_str)
414
376
 
415
377
    # Get extra trees that ought to be searched for file-ids
416
378
    extra_trees = None
417
379
    if working_tree is not None and working_tree not in (old_tree, new_tree):
418
380
        extra_trees = (working_tree,)
419
 
    return (old_tree, new_tree, old_branch, new_branch,
420
 
            specific_files, extra_trees)
421
 
 
 
381
    return old_tree, new_tree, specific_files, extra_trees
422
382
 
423
383
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
424
384
    if branch is None and tree is not None:
439
399
                    old_label='a/', new_label='b/',
440
400
                    extra_trees=None,
441
401
                    path_encoding='utf8',
442
 
                    using=None,
443
 
                    format_cls=None,
444
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
402
                    using=None):
445
403
    """Show in text form the changes from one tree to another.
446
404
 
447
 
    :param to_file: The output stream.
448
 
    :param specific_files: Include only changes to these files - None for all
449
 
        changes.
450
 
    :param external_diff_options: If set, use an external GNU diff and pass
451
 
        these options.
452
 
    :param extra_trees: If set, more Trees to use for looking up file ids
453
 
    :param path_encoding: If set, the path will be encoded as specified,
454
 
        otherwise is supposed to be utf8
455
 
    :param format_cls: Formatter class (DiffTree subclass)
 
405
    to_file
 
406
        The output stream.
 
407
 
 
408
    specific_files
 
409
        Include only changes to these files - None for all changes.
 
410
 
 
411
    external_diff_options
 
412
        If set, use an external GNU diff and pass these options.
 
413
 
 
414
    extra_trees
 
415
        If set, more Trees to use for looking up file ids
 
416
 
 
417
    path_encoding
 
418
        If set, the path will be encoded as specified, otherwise is supposed
 
419
        to be utf8
456
420
    """
457
 
    if context is None:
458
 
        context = DEFAULT_CONTEXT_AMOUNT
459
 
    if format_cls is None:
460
 
        format_cls = DiffTree
461
 
    with old_tree.lock_read():
 
421
    old_tree.lock_read()
 
422
    try:
462
423
        if extra_trees is not None:
463
424
            for tree in extra_trees:
464
425
                tree.lock_read()
465
426
        new_tree.lock_read()
466
427
        try:
467
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
468
 
                                                   path_encoding,
469
 
                                                   external_diff_options,
470
 
                                                   old_label, new_label, using,
471
 
                                                   context_lines=context)
 
428
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
429
                                                 path_encoding,
 
430
                                                 external_diff_options,
 
431
                                                 old_label, new_label, using)
472
432
            return differ.show_diff(specific_files, extra_trees)
473
433
        finally:
474
434
            new_tree.unlock()
475
435
            if extra_trees is not None:
476
436
                for tree in extra_trees:
477
437
                    tree.unlock()
478
 
 
479
 
 
480
 
def _patch_header_date(tree, path):
 
438
    finally:
 
439
        old_tree.unlock()
 
440
 
 
441
 
 
442
def _patch_header_date(tree, file_id, path):
481
443
    """Returns a timestamp suitable for use in a patch header."""
482
 
    try:
483
 
        mtime = tree.get_file_mtime(path)
484
 
    except FileTimestampUnavailable:
485
 
        mtime = 0
 
444
    mtime = tree.get_file_mtime(file_id, path)
486
445
    return timestamp.format_patch_date(mtime)
487
446
 
488
447
 
489
448
def get_executable_change(old_is_x, new_is_x):
490
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
449
    descr = { True:"+x", False:"-x", None:"??" }
491
450
    if old_is_x != new_is_x:
492
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
451
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
493
452
    else:
494
453
        return []
495
454
 
526
485
                     diff_tree.to_file, diff_tree.path_encoding)
527
486
 
528
487
    @staticmethod
529
 
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
 
488
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
530
489
        for file_differ in differs:
531
 
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
 
490
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
491
                                      new_kind)
532
492
            if result is not DiffPath.CANNOT_DIFF:
533
493
                return result
534
494
        else:
541
501
    Represents kind change as deletion + creation.  Uses the other differs
542
502
    to do this.
543
503
    """
544
 
 
545
504
    def __init__(self, differs):
546
505
        self.differs = differs
547
506
 
552
511
    def from_diff_tree(klass, diff_tree):
553
512
        return klass(diff_tree.differs)
554
513
 
555
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
514
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
556
515
        """Perform comparison
557
516
 
 
517
        :param file_id: The file_id of the file to compare
558
518
        :param old_path: Path of the file in the old tree
559
519
        :param new_path: Path of the file in the new tree
560
520
        :param old_kind: Old file-kind of the file
562
522
        """
563
523
        if None in (old_kind, new_kind):
564
524
            return DiffPath.CANNOT_DIFF
565
 
        result = DiffPath._diff_many(
566
 
            self.differs, old_path, new_path, old_kind, None)
 
525
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
526
                                       new_path, old_kind, None)
567
527
        if result is DiffPath.CANNOT_DIFF:
568
528
            return result
569
 
        return DiffPath._diff_many(
570
 
            self.differs, old_path, new_path, None, new_kind)
 
529
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
530
                                     None, new_kind)
571
531
 
572
532
 
573
533
class DiffDirectory(DiffPath):
574
534
 
575
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
535
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
576
536
        """Perform comparison between two directories.  (dummy)
577
537
 
578
538
        """
587
547
 
588
548
class DiffSymlink(DiffPath):
589
549
 
590
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
550
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
591
551
        """Perform comparison between two symlinks
592
552
 
 
553
        :param file_id: The file_id of the file to compare
593
554
        :param old_path: Path of the file in the old tree
594
555
        :param new_path: Path of the file in the new tree
595
556
        :param old_kind: Old file-kind of the file
598
559
        if 'symlink' not in (old_kind, new_kind):
599
560
            return self.CANNOT_DIFF
600
561
        if old_kind == 'symlink':
601
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
562
            old_target = self.old_tree.get_symlink_target(file_id)
602
563
        elif old_kind is None:
603
564
            old_target = None
604
565
        else:
605
566
            return self.CANNOT_DIFF
606
567
        if new_kind == 'symlink':
607
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
568
            new_target = self.new_tree.get_symlink_target(file_id)
608
569
        elif new_kind is None:
609
570
            new_target = None
610
571
        else:
613
574
 
614
575
    def diff_symlink(self, old_target, new_target):
615
576
        if old_target is None:
616
 
            self.to_file.write(b'=== target is \'%s\'\n' %
617
 
                               new_target.encode(self.path_encoding, 'replace'))
 
577
            self.to_file.write('=== target is %r\n' % new_target)
618
578
        elif new_target is None:
619
 
            self.to_file.write(b'=== target was \'%s\'\n' %
620
 
                               old_target.encode(self.path_encoding, 'replace'))
 
579
            self.to_file.write('=== target was %r\n' % old_target)
621
580
        else:
622
 
            self.to_file.write(b'=== target changed \'%s\' => \'%s\'\n' %
623
 
                               (old_target.encode(self.path_encoding, 'replace'),
624
 
                                new_target.encode(self.path_encoding, 'replace')))
 
581
            self.to_file.write('=== target changed %r => %r\n' %
 
582
                              (old_target, new_target))
625
583
        return self.CHANGED
626
584
 
627
585
 
632
590
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
633
591
 
634
592
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
635
 
                 old_label='', new_label='', text_differ=internal_diff,
636
 
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
 
593
                 old_label='', new_label='', text_differ=internal_diff):
637
594
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
638
595
        self.text_differ = text_differ
639
596
        self.old_label = old_label
640
597
        self.new_label = new_label
641
598
        self.path_encoding = path_encoding
642
 
        self.context_lines = context_lines
643
599
 
644
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
600
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
645
601
        """Compare two files in unified diff format
646
602
 
 
603
        :param file_id: The file_id of the file to compare
647
604
        :param old_path: Path of the file in the old tree
648
605
        :param new_path: Path of the file in the new tree
649
606
        :param old_kind: Old file-kind of the file
651
608
        """
652
609
        if 'file' not in (old_kind, new_kind):
653
610
            return self.CANNOT_DIFF
 
611
        from_file_id = to_file_id = file_id
654
612
        if old_kind == 'file':
655
 
            old_date = _patch_header_date(self.old_tree, old_path)
 
613
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
656
614
        elif old_kind is None:
657
615
            old_date = self.EPOCH_DATE
 
616
            from_file_id = None
658
617
        else:
659
618
            return self.CANNOT_DIFF
660
619
        if new_kind == 'file':
661
 
            new_date = _patch_header_date(self.new_tree, new_path)
 
620
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
662
621
        elif new_kind is None:
663
622
            new_date = self.EPOCH_DATE
 
623
            to_file_id = None
664
624
        else:
665
625
            return self.CANNOT_DIFF
666
 
        from_label = '%s%s\t%s' % (self.old_label, old_path,
667
 
                                   old_date)
668
 
        to_label = '%s%s\t%s' % (self.new_label, new_path,
669
 
                                 new_date)
670
 
        return self.diff_text(old_path, new_path, from_label, to_label)
 
626
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
627
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
628
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
629
            old_path, new_path)
671
630
 
672
 
    def diff_text(self, from_path, to_path, from_label, to_label):
 
631
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
632
        from_path=None, to_path=None):
673
633
        """Diff the content of given files in two trees
674
634
 
675
 
        :param from_path: The path in the from tree. If None,
 
635
        :param from_file_id: The id of the file in the from tree.  If None,
676
636
            the file is not present in the from tree.
677
 
        :param to_path: The path in the to tree. This may refer
678
 
            to a different file from from_path.  If None,
 
637
        :param to_file_id: The id of the file in the to tree.  This may refer
 
638
            to a different file from from_file_id.  If None,
679
639
            the file is not present in the to tree.
 
640
        :param from_path: The path in the from tree or None if unknown.
 
641
        :param to_path: The path in the to tree or None if unknown.
680
642
        """
681
 
        def _get_text(tree, path):
682
 
            if path is None:
683
 
                return []
684
 
            try:
685
 
                return tree.get_file_lines(path)
686
 
            except errors.NoSuchFile:
 
643
        def _get_text(tree, file_id, path):
 
644
            if file_id is not None:
 
645
                return tree.get_file(file_id, path).readlines()
 
646
            else:
687
647
                return []
688
648
        try:
689
 
            from_text = _get_text(self.old_tree, from_path)
690
 
            to_text = _get_text(self.new_tree, to_path)
 
649
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
650
            to_text = _get_text(self.new_tree, to_file_id, to_path)
691
651
            self.text_differ(from_label, from_text, to_label, to_text,
692
 
                             self.to_file, path_encoding=self.path_encoding,
693
 
                             context_lines=self.context_lines)
 
652
                             self.to_file)
694
653
        except errors.BinaryFile:
695
654
            self.to_file.write(
696
 
                ("Binary files %s and %s differ\n" %
697
 
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
655
                  ("Binary files %s and %s differ\n" %
 
656
                  (from_label, to_label)).encode(self.path_encoding))
698
657
        return self.CHANGED
699
658
 
700
659
 
704
663
                 path_encoding='utf-8'):
705
664
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
706
665
        self.command_template = command_template
707
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
666
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
708
667
 
709
668
    @classmethod
710
669
    def from_string(klass, command_string, old_tree, new_tree, to_file,
711
670
                    path_encoding='utf-8'):
712
 
        command_template = cmdline.split(command_string)
713
 
        if '@' not in command_string:
714
 
            command_template.extend(['@old_path', '@new_path'])
 
671
        command_template = commands.shlex_split_unicode(command_string)
 
672
        command_template.extend(['%(old_path)s', '%(new_path)s'])
715
673
        return klass(command_template, old_tree, new_tree, to_file,
716
674
                     path_encoding)
717
675
 
718
676
    @classmethod
719
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
677
    def make_from_diff_tree(klass, command_string):
720
678
        def from_diff_tree(diff_tree):
721
 
            full_command_string = [command_string]
722
 
            if external_diff_options is not None:
723
 
                full_command_string += ' ' + external_diff_options
724
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
679
            return klass.from_string(command_string, diff_tree.old_tree,
725
680
                                     diff_tree.new_tree, diff_tree.to_file)
726
681
        return from_diff_tree
727
682
 
728
683
    def _get_command(self, old_path, new_path):
729
684
        my_map = {'old_path': old_path, 'new_path': new_path}
730
 
        command = [AtTemplate(t).substitute(my_map) for t in
731
 
                   self.command_template]
732
 
        if sys.platform == 'win32':  # Popen doesn't accept unicode on win32
733
 
            command_encoded = []
734
 
            for c in command:
735
 
                if isinstance(c, text_type):
736
 
                    command_encoded.append(c.encode('mbcs'))
737
 
                else:
738
 
                    command_encoded.append(c)
739
 
            return command_encoded
740
 
        else:
741
 
            return command
 
685
        return [t % my_map for t in self.command_template]
742
686
 
743
687
    def _execute(self, old_path, new_path):
744
688
        command = self._get_command(old_path, new_path)
745
689
        try:
746
690
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
747
691
                                    cwd=self._root)
748
 
        except OSError as e:
 
692
        except OSError, e:
749
693
            if e.errno == errno.ENOENT:
750
694
                raise errors.ExecutableMissing(command[0])
751
695
            else:
752
696
                raise
753
697
        self.to_file.write(proc.stdout.read())
754
 
        proc.stdout.close()
755
698
        return proc.wait()
756
699
 
757
700
    def _try_symlink_root(self, tree, prefix):
758
 
        if (getattr(tree, 'abspath', None) is None or
759
 
                not osutils.host_os_dereferences_symlinks()):
 
701
        if (getattr(tree, 'abspath', None) is None
 
702
            or not osutils.host_os_dereferences_symlinks()):
760
703
            return False
761
704
        try:
762
705
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
763
 
        except OSError as e:
 
706
        except OSError, e:
764
707
            if e.errno != errno.EEXIST:
765
708
                raise
766
709
        return True
767
710
 
768
 
    @staticmethod
769
 
    def _fenc():
770
 
        """Returns safe encoding for passing file path to diff tool"""
771
 
        if sys.platform == 'win32':
772
 
            return 'mbcs'
773
 
        else:
774
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
775
 
            # handle utf-8 correctly when locale is not utf-8.
776
 
            return sys.getfilesystemencoding() or 'ascii'
777
 
 
778
 
    def _is_safepath(self, path):
779
 
        """Return true if `path` may be able to pass to subprocess."""
780
 
        fenc = self._fenc()
781
 
        try:
782
 
            return path == path.encode(fenc).decode(fenc)
783
 
        except UnicodeError:
784
 
            return False
785
 
 
786
 
    def _safe_filename(self, prefix, relpath):
787
 
        """Replace unsafe character in `relpath` then join `self._root`,
788
 
        `prefix` and `relpath`."""
789
 
        fenc = self._fenc()
790
 
        # encoded_str.replace('?', '_') may break multibyte char.
791
 
        # So we should encode, decode, then replace(u'?', u'_')
792
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
793
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
794
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
795
 
 
796
 
    def _write_file(self, relpath, tree, prefix, force_temp=False,
797
 
                    allow_write=False):
798
 
        if not force_temp and isinstance(tree, WorkingTree):
799
 
            full_path = tree.abspath(relpath)
800
 
            if self._is_safepath(full_path):
801
 
                return full_path
802
 
 
803
 
        full_path = self._safe_filename(prefix, relpath)
804
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
711
    def _write_file(self, file_id, tree, prefix, relpath):
 
712
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
713
        if self._try_symlink_root(tree, prefix):
805
714
            return full_path
806
715
        parent_dir = osutils.dirname(full_path)
807
716
        try:
808
717
            os.makedirs(parent_dir)
809
 
        except OSError as e:
 
718
        except OSError, e:
810
719
            if e.errno != errno.EEXIST:
811
720
                raise
812
 
        source = tree.get_file(relpath)
 
721
        source = tree.get_file(file_id, relpath)
813
722
        try:
814
 
            with open(full_path, 'wb') as target:
 
723
            target = open(full_path, 'wb')
 
724
            try:
815
725
                osutils.pumpfile(source, target)
 
726
            finally:
 
727
                target.close()
816
728
        finally:
817
729
            source.close()
818
 
        try:
819
 
            mtime = tree.get_file_mtime(relpath)
820
 
        except FileTimestampUnavailable:
821
 
            pass
822
 
        else:
823
 
            os.utime(full_path, (mtime, mtime))
824
 
        if not allow_write:
825
 
            osutils.make_readonly(full_path)
 
730
        osutils.make_readonly(full_path)
 
731
        mtime = tree.get_file_mtime(file_id)
 
732
        os.utime(full_path, (mtime, mtime))
826
733
        return full_path
827
734
 
828
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
829
 
                       allow_write_new=False):
830
 
        old_disk_path = self._write_file(
831
 
            old_path, self.old_tree, 'old', force_temp)
832
 
        new_disk_path = self._write_file(
833
 
            new_path, self.new_tree, 'new', force_temp,
834
 
            allow_write=allow_write_new)
 
735
    def _prepare_files(self, file_id, old_path, new_path):
 
736
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
737
                                         old_path)
 
738
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
739
                                         new_path)
835
740
        return old_disk_path, new_disk_path
836
741
 
837
742
    def finish(self):
838
743
        try:
839
744
            osutils.rmtree(self._root)
840
 
        except OSError as e:
 
745
        except OSError, e:
841
746
            if e.errno != errno.ENOENT:
842
747
                mutter("The temporary directory \"%s\" was not "
843
 
                       "cleanly removed: %s." % (self._root, e))
 
748
                        "cleanly removed: %s." % (self._root, e))
844
749
 
845
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
750
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
846
751
        if (old_kind, new_kind) != ('file', 'file'):
847
752
            return DiffPath.CANNOT_DIFF
848
 
        (old_disk_path, new_disk_path) = self._prepare_files(
849
 
            old_path, new_path)
850
 
        self._execute(old_disk_path, new_disk_path)
851
 
 
852
 
    def edit_file(self, old_path, new_path):
853
 
        """Use this tool to edit a file.
854
 
 
855
 
        A temporary copy will be edited, and the new contents will be
856
 
        returned.
857
 
 
858
 
        :return: The new contents of the file.
859
 
        """
860
 
        old_abs_path, new_abs_path = self._prepare_files(
861
 
            old_path, new_path, allow_write_new=True, force_temp=True)
862
 
        command = self._get_command(old_abs_path, new_abs_path)
863
 
        subprocess.call(command, cwd=self._root)
864
 
        with open(new_abs_path, 'rb') as new_file:
865
 
            return new_file.read()
 
753
        self._prepare_files(file_id, old_path, new_path)
 
754
        self._execute(osutils.pathjoin('old', old_path),
 
755
                      osutils.pathjoin('new', new_path))
866
756
 
867
757
 
868
758
class DiffTree(object):
896
786
            DiffPaths"""
897
787
        if diff_text is None:
898
788
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
899
 
                                 '', '', internal_diff)
 
789
                                 '', '',  internal_diff)
900
790
        self.old_tree = old_tree
901
791
        self.new_tree = new_tree
902
792
        self.to_file = to_file
910
800
    @classmethod
911
801
    def from_trees_options(klass, old_tree, new_tree, to_file,
912
802
                           path_encoding, external_diff_options, old_label,
913
 
                           new_label, using, context_lines):
 
803
                           new_label, using):
914
804
        """Factory for producing a DiffTree.
915
805
 
916
806
        Designed to accept options used by show_diff_trees.
917
 
 
918
807
        :param old_tree: The tree to show as old in the comparison
919
808
        :param new_tree: The tree to show as new in the comparison
920
809
        :param to_file: File to write comparisons to
926
815
        :param using: Commandline to use to invoke an external diff tool
927
816
        """
928
817
        if using is not None:
929
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
930
 
                using, external_diff_options)]
 
818
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
931
819
        else:
932
820
            extra_factories = []
933
821
        if external_diff_options:
934
822
            opts = external_diff_options.split()
935
 
 
936
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
937
 
                """:param path_encoding: not used but required
938
 
                        to match the signature of internal_diff.
939
 
                """
 
823
            def diff_file(olab, olines, nlab, nlines, to_file):
940
824
                external_diff(olab, olines, nlab, nlines, to_file, opts)
941
825
        else:
942
826
            diff_file = internal_diff
943
827
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
944
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
828
                             old_label, new_label, diff_file)
945
829
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
946
830
                     extra_factories)
947
831
 
948
832
    def show_diff(self, specific_files, extra_trees=None):
949
833
        """Write tree diff to self.to_file
950
834
 
951
 
        :param specific_files: the specific files to compare (recursive)
 
835
        :param sepecific_files: the specific files to compare (recursive)
952
836
        :param extra_trees: extra trees to use for mapping paths to file_ids
953
837
        """
954
838
        try:
961
845
        # TODO: Generation of pseudo-diffs for added/deleted files could
962
846
        # be usefully made into a much faster special case.
963
847
        iterator = self.new_tree.iter_changes(self.old_tree,
964
 
                                              specific_files=specific_files,
965
 
                                              extra_trees=extra_trees,
966
 
                                              require_versioned=True)
 
848
                                               specific_files=specific_files,
 
849
                                               extra_trees=extra_trees,
 
850
                                               require_versioned=True)
967
851
        has_changes = 0
968
 
 
969
852
        def changes_key(change):
970
853
            old_path, new_path = change[1]
971
854
            path = new_path
972
855
            if path is None:
973
856
                path = old_path
974
857
            return path
975
 
 
976
858
        def get_encoded_path(path):
977
859
            if path is not None:
978
860
                return path.encode(self.path_encoding, "replace")
990
872
            renamed = (parent[0], name[0]) != (parent[1], name[1])
991
873
 
992
874
            properties_changed = []
993
 
            properties_changed.extend(
994
 
                get_executable_change(executable[0], executable[1]))
 
875
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
995
876
 
996
877
            if properties_changed:
997
 
                prop_str = b" (properties changed: %s)" % (
998
 
                    b", ".join(properties_changed),)
 
878
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
999
879
            else:
1000
 
                prop_str = b""
 
880
                prop_str = ""
1001
881
 
1002
882
            if (old_present, new_present) == (True, False):
1003
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1004
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
883
                self.to_file.write("=== removed %s '%s'\n" %
 
884
                                   (kind[0], oldpath_encoded))
1005
885
                newpath = oldpath
1006
886
            elif (old_present, new_present) == (False, True):
1007
 
                self.to_file.write(b"=== added %s '%s'\n" %
1008
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
887
                self.to_file.write("=== added %s '%s'\n" %
 
888
                                   (kind[1], newpath_encoded))
1009
889
                oldpath = newpath
1010
890
            elif renamed:
1011
 
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
1012
 
                                   (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
 
891
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
892
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1013
893
            else:
1014
894
                # if it was produced by iter_changes, it must be
1015
895
                # modified *somehow*, either content or execute bit.
1016
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
1017
 
                                                                  newpath_encoded, prop_str))
 
896
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
897
                                   newpath_encoded, prop_str))
1018
898
            if changed_content:
1019
 
                self._diff(oldpath, newpath, kind[0], kind[1])
 
899
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1020
900
                has_changes = 1
1021
901
            if renamed:
1022
902
                has_changes = 1
1023
903
        return has_changes
1024
904
 
1025
 
    def diff(self, old_path, new_path):
 
905
    def diff(self, file_id, old_path, new_path):
1026
906
        """Perform a diff of a single file
1027
907
 
 
908
        :param file_id: file-id of the file
1028
909
        :param old_path: The path of the file in the old tree
1029
910
        :param new_path: The path of the file in the new tree
1030
911
        """
1031
 
        if old_path is None:
 
912
        try:
 
913
            old_kind = self.old_tree.kind(file_id)
 
914
        except (errors.NoSuchId, errors.NoSuchFile):
1032
915
            old_kind = None
1033
 
        else:
1034
 
            old_kind = self.old_tree.kind(old_path)
1035
 
        if new_path is None:
 
916
        try:
 
917
            new_kind = self.new_tree.kind(file_id)
 
918
        except (errors.NoSuchId, errors.NoSuchFile):
1036
919
            new_kind = None
1037
 
        else:
1038
 
            new_kind = self.new_tree.kind(new_path)
1039
 
        self._diff(old_path, new_path, old_kind, new_kind)
1040
 
 
1041
 
    def _diff(self, old_path, new_path, old_kind, new_kind):
1042
 
        result = DiffPath._diff_many(
1043
 
            self.differs, old_path, new_path, old_kind, new_kind)
 
920
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
921
 
 
922
 
 
923
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
924
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
925
                                       new_path, old_kind, new_kind)
1044
926
        if result is DiffPath.CANNOT_DIFF:
1045
927
            error_path = new_path
1046
928
            if error_path is None:
1047
929
                error_path = old_path
1048
930
            raise errors.NoDiffFound(error_path)
1049
 
 
1050
 
 
1051
 
format_registry = Registry()
1052
 
format_registry.register('default', DiffTree)